Research On Image Registration And Application

Image registration is an important issue in image procession, which is the first step for image fusion. Image registration technique has been used in many fields, which is basic component of medical image analysis, remote sensing image processing and target recognition. With the continuous developing of new sensor technology, the ability to obtain images with different physical characteristics is rapidly increased. So the multi-modality image registration technique becomes more and more important in research areas. People hope to reduce data errors through the registration of image acquired by different sensors, in order to achieve the purpose of increasing recognition rate and accuracy.In this paper, multi-modality image registration based on mutual information and optimization is the core of the experiment. Discuss the two important step of image registration, and propose improvements base on the summing up the comparison of advantages and disadvantages. To sum up the main work of this paper is as follows.(1)Describes the basics of image registration, including the concept of image registration and the application of image registration, and expound the four processes of image registration.(2)The optimization algorithms and measure function is described in detail, which are the two most important steps in image registration process, directly related to image registration accuracy and efficiency. It is concluded that the optimization algorithm will affect the registration result by comparing the three classical optimization algorithm;And then introduce three commonly used measure function, the experimental results show that: multi-modality image registration based on mutual information get ideal results.(3)A new image registration technique based on block mutual information and quantum-behaved particle swarm optimization algorithm is proposed. In the process of registration, the block mutual information is used for the similar measure, and the transformation parameters are calculated by using quantum-behaved particle swarm optimization algorithm. Experimental results show that the proposed method is efficient and it can avoid local minimum occurred in multimodality image registration, and gets ideal results.(4)2D/3D image registration method based on QPSO is described. This method is applied to the registration of X-ray fluoroscopy (simulated by DRR) and CT. In the experiment, with mutual information for measure, quantum-behaved particle swarm optimization is used to search for the 2D/3D image registration parameters. Experimental result shows that 2D/3D image registration based on quantum-behaved particle swarm of gets ideal results.(5)Lastly, the lack of the research in the paper is summarized,and then looks forward to the futurity of the research of image registration.